David R. Henderson  

First-Order Effects Are First

Where Superforecasters Start... Reply to Arnold Kling...

Co-blogger Bryan Caplan's excellent post on forecasting this morning reminds me of a similar point that co-author Charley Hooper (CLH) and I made in our book, Making Great Decisions in Business and Life.

Here it is:

First Find First-Order Effects

A decade ago I (CLH) had a conversation with a coworker about an upcoming football game between the Oakland Raiders--temporarily based in Los Angeles--and the Dallas Cowboys. He and I both shared an interest in the Raiders, but he loved to analyze football games in more detail than I thought was warranted.

Coworker: The Raiders should win this game because their defense is good and this game will be played on wet Astroturf. Players won't be able to get good footing in the cold and wet. The Raiders' defense will capitalize on this.
Me: The Raiders are 3-4 (three wins and four losses) and the Cowboys are 5-1. The Cowboys have won the last three match ups. The Raiders have a number of injured
players. I think the Cowboys will win.

As many as 100 factors could determine the ultimate winner of an NFL game. He seemed to focus on the less important ones and ignore those most important. His points were valid, but were they significant? By the way, the Raiders did lose 28-13.

Although I didn't follow football as much as he did, I predicted better because I focused on what was important.

I (CLH) have had conversations with people in the pharmaceutical industry that go something like this:

Them: This drug should do really well because we will have the full sales force pushing it.
Me: Yes, but this product's efficacy is low and doctors will fear its possible side effects.

What do this and the Raiders conversation have in common? They were both with intelligent people who made valid points to support their conclusions. Unfortunately, while their points were valid, they were overridden by other, equally valid, but more important, points. The benefit the Raiders would get from wet Astroturf is a second-order effect, subsidiary to other more important first-order effects. A first-order effect was the weak Raiders team and the strong Cowboys team. Engineers and scientists know that first-order effects drive the day, and second order effects come into play only when the first-order effects balance each other. In other words, to predict whether the Raiders will win or whether the new medicine will succeed, we should evaluate first-order effects before we consider second-order effects. First consider what is important.

Comments and Sharing

COMMENTS (2 to date)
_NL writes:

I think a further commonality is that both the superfan and the pharma employee are immersed in the thing they are analyzing, which encourages an emphasis on details.

People who are immersed in something prefer to think that their effort and proximity is rewarded. Someone who analyzes turf and weather thinks that this extra analysis will pay off and be really important. Someone who works in or near pharmaceutical marketing thinks that extra marketing will pay off and is really important.

Having gone through a political campaign as a paid staffer, we did a lot of things, went to meetings, hired consultants, aired ads, dropped literature at thousands of houses, and went to forums and debates. In November, the two-party vote was almost exactly the same as two years previously (roughly 55-45), when two different candidates ran. Politics makes you think campaigning matters a lot, but ultimately it matters only within the context of the universe of eligible and likely voters.

ThomasH writes:

This sounds like "buy low, sell high" as a get rich quick scheme. How do you know that a win-loss record is a "First Order" effect and that a wet field is a "Second Order" effect?

Comments for this entry have been closed
Return to top